charles a-tedd p-warren a-lessons historical dam incidents-2011

7/29/2019 CHARLES a-TEDD P-WARREN a-Lessons Historical Dam Incidents-2011

1/167

Lessons from historical dam incidents

Project: SC080046/R1


2/167

The Environment Agency is the leading public bodyprotecting and improving the environment in England andWales.

Its our job to make sure that air, land and water are lookedafter by everyone in todays society, so that tomorrowsgenerations inherit a cleaner, healthier world.

Our work includes tackling flooding and pollution incidents,reducing industrys impacts on the environment, cleaning uprivers, coastal waters and contaminated land, andimproving wildlife habitats.

This report is the result of research commissioned by theEnvironment Agencys Evidence Directorate and funded bythe joint Environment Agency/Defra Flood and CoastalErosion Risk Management Research and DevelopmentProgramme.

Published by:Environment Agency, Horizon House, Deanery Road,Bristol, BS1 5AHwww.environment-agency.gov.uk

ISBN: 978-1-84911-232-1

Environment Agency August 2011

All rights reserved. This document may be reproducedwith prior permission of the Environment Agency.

The views and statements expressed in this report arethose of the author alone. The views or statementsexpressed in this publication do not necessarilyrepresent the views of the Environment Agency and the

Environment Agency cannot accept any responsibility forsuch views or statements.

Further copies of this report are available from:The Environment Agencys National Customer ContactCentre by emailing:[email protected] by telephoning 08708 506506.

Author(s):J Andrew Charles, BREPaul Tedd, BREAlan Warren, Halcrow Group Ltd

Dissemination Status:Publicly available

Keywords:Dam, reservoir, embankment, impoundment, incident,disaster, accident

Research Contractor:Halcrow Group Ltd.Burderop ParkSwindon

Wiltshire SN10 5BPTel. 01793 812479

Environment Agencys Project Manager:Emma MilnerPhoenix HouseGlobal AvenueLeedsLS11 8PG

Collaborator:Building Research Establishment

Project Number:SC080046

Product Code:SCHO0811BUBA-E-E

Evidence Report Lessons from hi storical dam inci dentsii
http://www.environment-agency.gov.uk/mailto:[email protected]:[email protected]://www.environment-agency.gov.uk/


3/167

Evidence at theEnvironment AgencyEvidence underpins the work of the Environment Agency. It provides an up-to-dateunderstanding of the world about us, helps us to develop tools and techniques tomonitor and manage our environment as efficiently and effectively as possible. It alsohelps us to understand how the environment is changing and to identify what the futurepressures may be.

The work of the Environment Agencys Evidence Directorate is a key ingredient in thepartnership between research, guidance and operations that enables the EnvironmentAgency to protect and restore our environment.

This report was produced by the Research, Monitoring and Innovation team withinEvidence. The team focuses on four main areas of activity:

Setting the agenda, by providing the evidence for decisions;

Maintaining scientific credibility, by ensuring that our programmes andprojects are fit for purpose and executed according to international standards;

Carrying out research, either by contracting it out to research organisationsand consultancies or by doing it ourselves;

Delivering information, advice, tools and techniques, by makingappropriate products available.

Miranda Kavanagh

Director of Evidence

Evidence Report Lessons from historical dam inci dents iii


4/167

Executive summaryThis report aims to help those responsible for the safety of reservoirs. These includeengineers appointed under current legislation, personnel who visit reservoirs in thecourse of their duties, staff who operate and monitor reservoirs, and enforcement

authority engineers.

The scope of the report is limited to water-retaining structures: some types of wasteimpoundments, such as tailings dams, may suffer similar types of malfunction, butthese are not included. Although the focus of the report is on incidents at dams in GreatBritain, reference is made to a few international incidents. International experience isparticularly helpful for those types of dam that are not commonly found in Great Britain.

The report begins with an introduction in Section 1. The background to the subject isbriefly outlined, the value of the national incident database is demonstrated and theneed for post-incident reporting and investigation is emphasised.

In the next two sections, general, technical and regulatory lessons from dam incidentsare outlined. Section 2 gives a historical overview of the subject which shows howserious incidents have improved our understanding of dam behaviour and the hazardsposed by these structures. This section should not only be of interest to dam engineersbut should also help those reservoir owners with limited technical knowledge todevelop a basic grasp of the more significant aspects of the subject. Section 3 showsthe close links between historical incidents and failures and the development ofreservoir safety legislation and guidance.

Section 4 looks at how incidents have been managed, including the role of owners andpanel engineers. The significance of drawdown rates and other provisions for damincidents such as evacuation planning are presented. Some examples of incidentmanagement are described.

Section 5 begins with an overview of serious incidents and a classification and briefanalysis of the modes of failure. This is followed by descriptions of over thirty majorincidents and summaries of seventy other incidents. There is some overlap withinformation presented in Section 2, but Section 5 in essence constitutes a convenientreference section for readers interested in incidents of a particular type or at aparticular dam.

Evidence Report Lessons from hi storical dam inci dentsiv


5/167

AcknowledgementsIn the preparation of this report by Halcrow/Building Research Establishment (BRE)substantial use has been made of work previously published on the subject:

Binnie (1976) Theevolution of British dams. Skempton (1989) Historical development of British embankment dams to 1960. Kennard (1995) Four decades of development of British embankment dams.

Work previously published by BRE has also been of assistance including Charles(1986) The significance of problems and remedial works at British earth dams, Charles(1990) Deterioration of clay barriers, and Tedd et al. (1994) Remedial works to claycores of UK embankment dams.

Mr J R Claydon undertook a peer review of the selected dam incidents and Dr A KHughes reviewed a draft report.

The authors would also like to thank the following who provided valuable comments onthe final draft:

A J BrownJ R ClaydonL DeucharD P M DuttonD M CrookC J FalkinghamC HoskinsP Kelham

K D GardinerR MannA C RobertshawA RowlandN Williams

Evidence Report Lessons from historical dam inci dents v


6/167

Contents

1 Introduction 1

1.1 Background 1

1.2 National incident database 4

1.3 Post-incident reporting 5

2 Technical lessons from dam incidents 8

2.1 Dam and reservoir failure 8

2.2 Nineteenth century 12

2.3 Early twentieth century (1901-1930) 16

2.4 Mid-twentieth century (1931-1960) 18

2.5 Late twentieth century (1961-2000) 21

2.6 Twenty-first century 25

2.7 Lessons from serious incidents 26

3 Dam incidents and the development of reservoir safety legislation 29

3.1 Nineteenth century 29

3.2 Twentieth century 31

3.3 Twenty-first century 34

4 Incident management 36

4.1 Provisions for managing incidents 364.2 Examples of dam incident management 37

5 Descrip tion of incidents 43

5.1 Introduction 43

5.2 Description of major incidents 56

5.3 Short summaries of additional incidents 118

References 148

Evidence Report Lessons from hi storical dam inci dentsvi


7/167

1 Introduction

1.1 BackgroundThe number of casualties arising from a breached dam can be greater than frommost other kinds of technological disaster. Maintaining reservoir safety hasconsiderable importance for the public in a country such as Great Britain where anumber of dams pose a high hazard, being located upstream of heavily populatedand industrialised areas. Thus, although the probability of failure of a dam isgenerally low, the consequences of failure could be great. As most reservoirsconstitute a low probability/high consequence scenario, careful management of theserisks is essential.

Fortunately, few catastrophic failures have occurred in Great Britain and since 1925

there has been no loss of life due to dam disasters. Table 1.1 lists dam failures thatcaused loss of life in Great Britain. All the dams are embankments except Eigiauwhich was concrete and failed due to an inadequate foundation. Since 1925, therehave been failures involving breaching of embankments and also many near missesand other serious incidents(Wright, 1994).

Table 1-1 British dam failures that caused loss of life (after Charles, 1993)Failure

DamH

(m)

Reservoirvolume

(x 103m

3)

Datebuilt

Date Type

Deaths

Tunnel End 9 1798 1799 OF 1

Diggle Moss (Black Moss) 1810 1810 OF 5Whinhill 12 262 1828 1835 IE 31Brent (Welsh Harp) 7 1837 1841 OF 2Glanderston 1842 OF 8Bold Venture (Darwen) 10 20 1844 1848 OF 12Bilberry 29 310 1845 1852 IE 81Dale Dyke 29 3,240 1863 1864 IE 244Rishton 1870 3Cwm Carne 12 90 1792 1875 OF 12Castle Malgwyn 1875 OF 2Clydach Vale 1910 OF 5Skelmorlie 5 24 1861 1925 OF 5Eigiau and 10 4,500 1911 1925 FF

Coedty (Dolgarrog) 11 320 1924 1925 OF16

Type of failure: IE =internal erosion, FF =foundation failure, OF =overtopping during flood

Although there has been no loss of life since 1925 due to dam disasters in GreatBritain, during the last fifty years disastrous failures overseas have resulted in muchloss of life as shown by the examples in Table 1.2.

Much can be learned from these failures, particularly those such as Baldwin Hills,Malpasset, Teton and Vaiont which have been the subject of detailed investigationand substantial literature. A useful start to such a study is given by J ansen in hisbook Dams and public safety (J ansen, 1980) which includes illuminating accounts ofthe failures of Machhu II, Teton, Frias, Baldwin Hills, Vaiont, Babii Yar, Malpasset,

and Vega de Tera as well as of many other failures. Failure of Vaiont and Malpasset

Evidence Report Lessons from hi storical dam inci dents 1


8/167

are described in further detail in Section 5 of this report. Both involved large loss oflife.

Table 1-2 Some international dam disasters causing loss of lifeFailureDam Dam

typeCountry Height

(m)

Res.volume(10

6m

3)

Datebuilt

Date Type

Deaths

Vega de Tera CMB Spain 34 7.8 1957 1959 SF 144

Malpasset CA France 66 22 1954 1959 FF 421

Babii Yar Emb Ukraine 1961 OF 145

Vaiont CA Italy 265 150 1960 1963 L 2,600

Baldwin Hills Emb USA 71 1.1 1951 1963 IE 5

Frias Emb Argentina 15 0.2 1940 1970 OF 42+

Banqiao Emb China 118 492 1953 1975 OF #

Teton Emb USA 93 308 1975 1976 IE 11

Machhu II Emb India 26 100 1972 1979 OF 2,000

Bagauda Emb Nigeria 20 0.7 1970 1988 OF 50Belci Emb Romania 18 13 1962 1991 OF 25Gouhou Emb China 71 3 1989 1993 IE 400

Zeizoun Emb Syria 42 71 1996 2002 OF 20Camara RCC Brazil 50 27 2002 2004 5Shakidor Emb Pakistan 2003 2005 OF 135+Situ Gintung Emb Indonesia 16 2 2009 IE 100

Dam type: CA =concrete arch, CMB =concrete and masonry buttress, Emb =embankment, RCC =roller compacted concrete.

Type of failure: IE =internal erosion, FF =foundation failure, OF =overtopping during flood, SF =structural failure on first filling, L =270 x 106 m3 landslide into the reservoir caused overtopping of thedam by a wave 125 m high, but remarkably the dam survived.#=It has been reported that tens of thousands died in this disaster which involved the failure of anumber of dams, of which Banqiao was the largest.

News items in New Civil Engineerwith headings such as Dam emergency ringschecking alarm bells (23 J anuary 2003) and Unstable dam assessed after 10 yearsneglect (30 J anuary 2003) show that near misses continue to occur. It was reportedinNew Civil Engineer(21 November 2002) that reservoir engineers told NCE lastweek that as many as four dams and reservoirs could be at risk from bursting everyyear".

Given the broad scope of this subject, the report has been limited in several ways:

Although reservoir safety is essentially an international subject, andworldwide experience is of considerable value, the report is for the most partlimited to dams and reservoirs in Great Britain. International incidents areincluded only to illustrate types of incident that are not covered by the Britishdatabase of dam incidents.

Although waste impoundment structures such as tailings dams can presentsimilar hazards to water-retaining dams, these are not included as separatesafety legislation is currently in force.

The Reservoirs Act 1975 applies to large raised reservoirs holding more than 25,000cubic metres of water, but this report includes information on smaller reservoirs

Evidence Report Lessons from hi storical dam inci dents2


9/167

since, depending on their location and elevation, smaller dams can present asubstantial threat to public safety and lessons can be learned from incidents at suchreservoirs. Furthermore, such reservoirs are likely to be brought within the ambit ofnew legislation.

There are several approaches to mitigating the risk and consequences of a dam

failure. The risk of failure may be reduced by structural improvements to the dam andits ancillary works and by better surveillance, monitoring and maintenance. Tighteremergency management procedures can reduce the likelihood of failure and risk ofcasualties should a failure occur.

The majority of British reservoirs are impounded by embankment dams, many ofthem built in the nineteenth century (BRE, 1994). Considering the emphasis given toslope stability in geotechnical engineering, it may seem surprising that relatively fewcatastrophic failures have been due to slope instability associated with inadequateshear strength or high pore pressures. Most of the failures which have caused loss oflife can be attributed to the embankment breaching due to one of two causes:

Overtopping of the embankment during an extreme flood. This hazard islargely within the province of hydrology and the selection and estimation ofthe design flood, and provision of appropriately sized spillway and freeboard.

Internal erosion associated with processes such as piping or hydraulicfracture. In new dams this should be prevented by appropriately designedfilters and careful design of the watertight element. Where overflowarrangements have been improved to meet modern flood standards, internalerosion is likely to be the major remaining threat to an old embankment damwhich does not have filters designed to modern standards or which has adraw-off structure (culvert or unprotected pipe) passing through it or whichhas a deep clay filled cut-off trench.

Knowledge of the dam and its ancillary works, and of processes likely to be at workwhich could pose a threat to safety, make it possible to assess the hazard of internalerosion Charles (1998, 2002a). An understanding of the performance of similar damsmay become more critical as the stock of dams in the United Kingdom ages. Theeffects of climate change, changes in operating conditions and the ageing processitself might change the patterns of geotechnical behaviour understood from historicalperformance and incidents. Learning from the recorded performance of dams isfundamental to improving reservoir safety and, consequently, incident reporting andthe compiling of case histories are important tasks. Until recently, this was done onanad hoc basis by publishing case histories, but many incidents remain unreported.Nevertheless, major failures have generally been reported and discussed in learned

journals or conferences.

Since 1930, reservoir safety in Great Britain has been regulated by Act of Parliament.In the interests of public safety, the Reservoirs (Safety Provisions) Act 1930 requiredthe owners of reservoirs with a capacity of more than five million gallons (22,700 m3)above the natural level of any part of the surrounding ground, to provide for theirinspection by a qualified civil engineer who was a member of a panel of civilengineers constituted for the purposes of the Act. The Reservoirs Act 1975 wentbeyond the provisions of the earlier Act in a number of ways. Local authorities weredesignated as enforcement authorities, being required to keep registers of all raisedreservoirs (defined as those with a capacity greater than 25,000 m3 above the natural

level of any part of the land adjoining the reservoir) and to ensure that undertakers,usually the owners, complied with the requirements of the Act. The duties of



10/167

undertakers, enforcement authorities and engineers appointed to the various panelswere laid down in the Act or set out in regulations.

A major change in reservoir safety occurred in September 2004 when responsibilityfor the enforcement of safety legislation in England and Wales was transferred from alarge number of local authorities to the Environment Agency under the provisions ofthe Water Act 2003, thereby ensuring a uniform application of safety legislationacross the country. The Flood Risk Management (Scotland) Act 2009 transfers theEnforcement Authority role to SEPA. Further legislative changes are planned in theFlood & Water Management Act 2010.

1.2 National incident database

The study of case histories has a major role in subjects as diverse as medicine, lawand engineering design. While it is vital that practitioners have a sound grasp of theunderlying principles of their subject, their personal practical experience needs to be

supplemented by the study of well-documented case studies. Although the teachingof engineering science is primarily concerned with analysis, case histories shouldplay a role in engineering education, particularly for civil and geotechnical engineerswho assess the condition of existing works.

Knowledge of the history of a dam is one of the most useful and important elementsin making an accurate diagnosis of a reservoir safety problem and in some cases canbe more valuable than physical examinations and diagnostic tests. It should includethe records of monitoring and surveillance, previous incidents and remedial works.Case histories have useful functions in a number of areas:

(a) In dealing with an emergency, readily available documentation of thehistory of the dam can be a crucial factor.

(b) In the condition assessment of a dam, knowledge of past behaviour of thistype of structure is important and should help to identify abnormal behaviour.J ust as no competent physician would treat a patient without first ascertainingas much of the patients relevant medical history as possible, so no engineershould diagnose the nature of a problem or design remedial works at aparticular dam, without first researching the history of the structure.

(c) Where a particular type of problem has been diagnosed at a dam, a studyof the case histories of dams with similar problems and remedial works can

be a useful guide when considering the best course of action.

(d) A collection of case histories can give an indication of the prevalence orotherwise of different types of malfunction in a particular type of dam and canprovide a useful indication of the need for preventative works and of possiblesolutions. The Building Research Establishment (BRE) began developing anational dams database in 1988 as part of the Government's Reservoir SafetyResearch Programme and this included compiling data on dam failures andincidents, and remedial works (Tedd et al., 1992).

Needless to say, there are limitations and problems with the study of case histories.Only a fraction of the information potentially available is likely to be readily available.

The reliability of a case study involving a malfunction of a dam may be suspectbecause it is incomplete, or a full account might have been embarrassing for some of



11/167

the parties involved and professional ethics may forbid the exposure of details givenconfidentially. Published case studies are comparatively rare and may be untypical,and on occasion may have been carefully selected to prove a particular theory. TheBRE Bibliography of British Dams provides a comprehensive publicly available list ofpublished references (the bibliography can be accessed on the website of the BritishDam Society: www.britishdams.org). The long-term preservation of data presentsproblems as it is difficult to store data so that it is accessible and yet secure.

The BRE database was superseded by the new national incident database in 2006as part of work on the post-incident reporting system and this is now administered bythe Environment Agency under the guidance of an independent All Reservoirs PanelEngineer. The database holds information on dam characteristics and remedialworks as well as information on incidents. The database contains a substantialamount of incident data which has been useful in assessing the probability of a safetyincident at a British embankment dam (Brown and Tedd, 2003).The EnvironmentAgency also holds basic information on all statutory reservoirs in England and Walesand these provisions should provide a firm foundation for incident reporting and

identifying future research needs. The national incident database is freely available tothose with a legitimate need to access the information.

1.3 Post-incident reporting

An incident reporting system can be helpful in enhancing public safety in hazardoussituations and such systems have been developed in many high hazard industries(McQuaid, 2002). In the context of reservoir safety, an incident can be defined as anevent which differs from normal conditions and which has resulted in, or could havehad the potential to result in, an uncontrolled release of reservoir water, withconsequent harm to people, property or the natural environment. The most significantuncontrolled release of reservoir water is likely to be associated with the breaching ofa dam, but failure of ancillary works can also cause hazardous situations. Seriousincidents include:

(a) failures in which there has been an uncontrolled release of reservoir water withconsequent casualties or property damage;

(b) near misses which have not caused casualties or property damage, but whichmight have done had there been no human intervention; typically a near missincident requires emergency action such as rapid reservoir drawdown, theimplication being that without such emergency action a breach would be likely.

The Environment Agency reporting system covers gathering, analysing and sharinginformation about reservoir incidents. The system also provides for the investigationof the more serious, unusual or complex incidents, such as the Ulley incident in 2007.The system of incident reporting should help to identify and quantify trends in thebehaviour of dams subject to reservoir safety legislation and provide comprehensiveinformation on incidents that will help determine future research priorities(Charles,2005). The system aims to:

gather information on reservoir safety incidents;

investigate incidents where appropriate;

learn lessons from incidents; inform the reservoir industry of trends and lessons learned;



12/167

provide information that can contribute to reservoir safety research andincident frequency data for quantitative risk assessment.

Where there are particular points of learning that should be shared, bulletins areprepared to provide an insight into an incident or group of incidents. Each year, the

Environment Agency publishes an annual report on post-incident reporting to reviewthe incidents reported over the last year and to provide an update on relatedresearch and development.

Fortunately, failures are rare and a system of reporting that also includes nearmisses has much to commend it. Often, more can be learned from a near miss thanfrom a failure:

(a) When a failure occurs, matters of blame, legal responsibility and thepossibility of criminal prosecution can form a difficult environment in which tocarry out a satisfactoryinvestigation. The investigation of a near miss doesnot have these problems to the same extent.

(b) An uncontrolled release of reservoir water is generally associated with abreach of the dam and evidence of the cause of failure is likely to bedestroyed in the failure. With a near miss, the evidence still exists and canbe fully investigated.

(c) The much higher rate of near misses than failures facilitates meaningfulquantitative analysis and provides insight into the probability of failure fromdifferent causes. Reports of near misses provide a reminder of hazards andencourage timely preventative actions. Potential failures are identified beforean accident occurs.

(d) Reports of near misses help to identify the reasons why failures do notoccur. The report of an incident should reveal the barriers that prevented anear miss becoming a failure. This may shed light on whether near missesare a good guide to failures.

Where internal erosion takes place, the reason that the near miss did not become afailure may be associated with some or all of the following factors:

early identification of the problem;

rapid drawdown of the reservoir;

slow development of internal erosion.

A review of the Upper Rivington incident commissioned by the Department forEnvironment, Food and Rural Affairs (Defra) focused on safety legislation (for whichthe department has responsibility) and on the effectiveness of Defras ReservoirSafety Research Programme. The technical causes of the incident were notinvestigated specifically, but were not entirely excluded from the study. The Reviewof operation of Reservoirs Act 1975 in relation to serious incident at Upper Rivington(May 2002) was carried out with the cooperation of the reservoir undertakers, andmade the recommendations shown in Table 1.3 (Charles, 2005).

Table 1-3 Recommendations from report of serious incident in 2002

Recommendations Actions

The proposal to amend the Reservoirs Act to givethe Secretary of State powers to direct undertakers

Following the Water Act 2003,undertakers of specific reservoirs will



13/167

to prepare a plan setting out the action they wouldtake to control or mitigate the effects of floodinglikely to result from any escape of water from thereservoir should be implemented as soon aspossible.

be required to prepare reservoir floodplans. These plans must set out howundertaker will respond in anemergency to reduce the effects offlooding due to an escape of waterfrom the reservoir.

Helpful guidance on appropriate emergencyprocedures for rapid lowering of the reservoir in anemergency has been given inAn engineering guideto the safety of embankment dams in the UnitedKingdom. However, in view of the vital role suchprocedures have in maintaining reservoir safety,consideration should be given to whether furtherguidance is required to emphasise their importance.

A suite of papers has been publishedin Dams & Reservoirs on the use oflow-level outlets in emergencysituations. The papers deal with,respectively, target capacity(Hinks(2009), risk assessment (Brown(2009a) and British Waterwaysapproach (Brown, 2009b).

A formal system of reporting serious incidentsshould be developed. Investigations of those

incidents which might be termed a "near miss"would also be helpful. The requirements for incidentinvestigation in the nuclear industry, for chemicalhazards and for railways have been reviewed by theHealth and Safety Executive (2000).

A system for reporting seriousincidents has been developed

(Gosden and Brown, 2004). TheEnvironment Agency has taken thelead role and has developedprocesses and procedures for avoluntary system which wasimplemented in 2007 (Warren andHope, 2006).

Further research on internal erosion should beundertaken. A preliminary study should define thescope of the work to ensure its relevance to thethreat internal erosion poses to the Britishpopulation of old embankment dams.

This long-term objective is beingfacilitated through internationalcollaboration through the ICOLDEuropean Working Group on internalerosion (Charles, 2002a). An outline

strategy has been devised (Brownand Gosden, 2004).

This report provides many other examples of how serious incidents and near misseshave shaped reservoir safety legislation and best practice in dam design.



14/167

2 Technical lessons from damincidents

2.1 Dam and reservoir failure

Dams vary greatly in their design and construction. Their effectiveness and safetyalso depend critically on their foundation and all sites differ in their geology. Humanfactors affect how a reservoir is operated and how a dam is maintained, monitoredand kept under surveillance. With so many variables, recording and learning fromhow dam incidents have arisen is challenging. This section provides an overview ofthe types of major incidents that have arisen over the last 200 years.

If asked to cite failures of British dams, most engineers in the reservoir industrywould be able to quote Dale Dyke, Bilberry and Dolgarrog, together with recentserious incidents such as Ulley, but many would struggle to name more of theseveral hundred incidents that have occurred. The lack of knowledge of damincidents can give rise to misplaced optimism with respect to the long-termdeterioration of dams. This report aims to counter this by providing a broadperspective on the range of incidents that have arisen in the past and can ariseagain.

With most structural failures damage is limited to an area in the immediate vicinity ofthe structure, but the breaching of a dam and the consequent uncontrolled release ofthe impounded reservoir water can cause destruction over a large area downstream

of the dam. The structural stability and security of such dams, therefore, is of majorimportance for public safety, particularly in Great Britain where many reservoirs arelocated in river valleys upstream of densely populated and industrial areas.

There is a long history of dam and reservoir construction in Britain. In the second halfof the eighteenth century, many ornamental lakes were established in thelandscaped grounds of country estates and, by the end of the century, reservoirswere needed to supply the canals rapidly being built across the country. During thefirst half of the nineteenth century, the demand for unpolluted water supplies to therapidly expanding industrial towns led to a major increase in reservoir construction.This continued throughout the nineteenth and twentieth centuries, but has beenfollowed by a decline in dam construction during the last thirty years.

Figure 2.1 shows the growth in reservoir capacity and number of reservoirs since1800 (Tedd et al., 2000). It shows that older dams were generally low structures withsmall reservoir capacities and the effect of constructing much larger reservoirs in the1950s, in particular the construction of hydro-electric schemes in Scotland.

The timeline in Figure 2.2 shows important incidents and developments in damconstruction and legislation.

Before 1900, nearly all British dams were of the embankment type with a notableexception in Vyrnwy, a gravity dam built in 1890 to supply water to the city of

Liverpool. Although in the twentieth century a large proportion of dams were built ofmasonry or concrete, the majority of dams in Great Britain are earth embankments.Reservoir safety is thus intimately concerned with the behaviour and long-term



15/167

performance of old embankment dams. Until the 1950s, most embankment dams inBritain were built to a traditional design with a central core of puddle clay. Moremodern embankment dams typically use a wider core of rolled clay. However, theuse puddle clay cores dam continued until 1972, with the completion of J umbles,north east of Manchester.

0

2

4

6

8

1800 1850 1900 1950 2000

Date

Reservoir

capacity(109

m3)

0

10

20

30

40

50

60

70

80

90

100

Numberof

reservoirsbuiltas

percentageoftotal

Damconstruction

Reservoir

capacity

Figure 2-1 Growth in Briti sh reservoir capacity and number of reservoirs wi th time

The vast majority of serious incidents have concerned embankment dams, but this isnot surprising since about a tenth of British dams are built of concrete/masonry. Itcertainly should not be concluded that concrete dams are immune to problems.

While a dam failure can be broadly defined as an incident, occurrence or processwhereby a dam does not perform the function for which it has been constructed,namely to safely impound a reservoir of water, dam failure generally means thebreaching of a dam with the uncontrolled release of reservoir water. However, theterm is also used for the failure of an embankment dam during construction; that is,

an event where inadequate shear strength in the fill and possibly the foundationcauses instability in one or both embankment slopes and may also involve thefoundations. If such instability occurs before the embankment has begun to act as awater-retaining structure, there is less concern for reservoir safety.

When a dam has been constructed and the reservoir basin filled with water, the damis said to be in service. Failures in service are distinct to those arising during damconstruction. Unlike failures during construction, failures in service are usuallyintimately connected with the particular function of the dam to impound a reservoir ofwater. Shear failure could, of course, occur with the reservoir fully impounded as wellas during construction, but there are other more common modes of failure. If anembankment dam crest is overtopped during a flood, it could be breached by surface

erosion of floodwater flowing over the crest and downstream slope. Failure byovertopping during construction is rare and the only incident of this type is included in



16/167

Section 5 (Woodhead No1 overtopped on 12 October 1849 releasing 500 x 103 m3 ofwater). An embankment dam can also fail by internal erosion associated withexcessive seepage and leakage passing through the body of the embankment or thefoundation.

Some hazards, such as foundation instability as at Eigiau, threaten concrete dams aswell as embankment dams, whereas other hazards relate to concrete deterioration.Examples include long-term seepage, frost, the use of high-alumina cement andalkali-aggregate reaction.

The Environment Agency report Modes of dam failure and monitoring and measuringtechniques provides further information on the threats to dam safety and the variousways in which incidents might arise.



17/167

Si gn if ic an t B ri ti sh fai lu res an d i nc id en ts Dat e Dat e

Key developments in British dam construction practice, legislation

and guidance

1730 First use of clay core, Serpentine (Hyde Park)

1755 Use of clay blanket on upstream face, Petworth

1766 First foundation cut-off

1795 Butterley - one of first puddle clay core dams

Failure of Blackbrook due to poor construction 1799

Whinhill, 31 dead 1835

1838

Telford set standard for puddle caly core dam design with hydraulic gradient of 3

across the core

Failure of Bilberry, 81 dead 1852

1854-1862Slope failures during construction at Arnfield, Calf Hey and Piethorne repairedincorporating a toe berm.

1854-1864 Specifications for zoned fill construction become common practice.

Ainsworth Mill Lodge floods mine workings 1860 1868 Imposition of strict liability "Rylands v Fletcher"

Failure of Dale Dyke, 244 dead 1864 1864

Specification by Simpson (one of the investigating engineers) incorporates points

of learning: railway wagons to be excluded from embankment area during

construction; fill to be worked in horizontal layers not exceeding 9 inches.

1866

Introduction of Waterworks Bill with many of the features of the Reservoirs (Safety

Provisions) Act 1930

1872

Rankine's paper on gravity dam design published (principal stress and 'middle

third' concepts).

Pentwyn serious internal erosion in puddle clay cut-off 1870s 1876 Woodhead No. 2 dam, first use of concrete to completely fill deep cut-off trench.

1877

Binnie promotes the use of 'long inclines' rather than abrupt changes in the depth

of puddle trenches to reduce the risk of differential settlement.

Tunstall and Cowm, severe leakage through

foundation 1879 1879 First remedial use of grouting to seal foundations

1882

Vyrnwy gravity dam designed with a drainage tunnel network to reduce uplift

pressures.

Failure of Skelmorlie, 5 dead 1925

Dolgarrog failure, 16 dead 1925 1925

Edward Sandeman letter published in the Times which led to Reservoirs (Safety

Provisions) Act 1930.

1930 Reservoirs (Safety Provisions) Act 1930

1933Publication of "Interim report of the Committee on Floods in relation to ReservoirPractice"

Construction failure of Chingford 1937 1937

se o mo ern eart movng an constructon equpment e to g pore

pressures. Soil mechanics used in re-design for first timeof British dam.

Lynmouth flood disaster 1952 1960

Re-publication of "Interim report of the Committee on Floods in relation to

Reservoir 1960 Practice"

1960s

End of puddle clay core construction and beginning of rolled clay core construction

in the UK

Balderhead, severe internal erosion on first filling 1967 1968

First use of diaphragm walls to repair watertight element of dam. Development of

filter rules based on permeability of filter.

Lluest Wen; internal erosion emergency drawdown 1969Warmwithens; failure with uncontrolled release of

water. 1970

1975 Reservoirs Act 1975; publication of the 'Flood Studies Report'

1978 Publication of "Floods and Reservoir Safety: an engineering guide" 1st edition

Construction failure of Carsington 1984 1986

Instigation of Review Panels and the National Database at BRE following

recommendations in the Coxon Report into the failure of Carsington dam

1986 Reservoirs Act 1975 comes into full effect

1990

Publication of " An engineering guide to the safety of embankment dams in the

United Kingdom" 1st edition

Kielder; disruption to upstream blockwork 1984 1995 Publication of "Performance of blockwork and slabbing protection for dam faces"

1999 Publication of the "Flood Estimation Handbook"

Boltby; failure of stepped masonry spillway 2005

2007 Post-incident reporting system established

2007 EA Biennial Report sets out proposals for legislative change

Ulley; failure of spillway, emergency drawdown and

evacuation 2007 2008 The Pitt Report. Paper on "Security of stepped masonry spillways" published

2009

Draft 'Floods and Water Management Bill' sets out proposed changes to the

Reservoirs Act for England and Wales; Flood Risk Management (Scotland) Bill

sets out changes for Scotland.

Figure 2-2 Timeline of failures/incidents and key developments in Bri tish damconstruction practice, legislation and guidance



18/167

2.2 Nineteenth century

By the middle of the nineteenth century a fairly standard design of embankment damhad been adopted, with an upstream slope of one vertical in three horizontal and asteeper downstream slope of one vertical in 2.5 or two horizontal. Although control of

material and workmanship should ensure the integrity and watertightness of the corewithin the body of the dam, leakage could occur through the natural strata of thevalley underneath or around the sides of the dam. Leakage could also occur in thebasin of the reservoir. In the early puddle clay core dams, it was usual to extend thepuddle clay into a cut-off trench below ground level thus connecting the core to astratum of low permeability. The trench often continued into the valley sides.Sometimes very deep trenches were dug but the trench was usually narrow, withvertical sides.

The catastrophic failure of two dams in the nineteenth century, Bilberry in 1852 andDale Dyke in 1864, led to major changes in the design and construction of puddleclay core embankment dams.

Excavation for the cut-off trench of the 29-metre high Bilberry dam began in 1839and a spring was encountered in the bottom of the trench. The outlet workscomprised a masonry culvert which had to cross the puddle clay filled cut-off trench.Serious problems soon became apparent: muddy water came through the culvert in1841 and in 1843 the leakage became worse and water burst through the culvert.Remedial works were unsuccessful and large settlements occurred. It was claimedthat between 1846 and 1851 the bank settled three metres. This settlementeliminated the freeboard and soon after midnight on 5 February 1852 theembankment was overtopped and breached during a storm. The resulting floodclaimed 81 lives in the Holme Valley below the dam. It would appear that erosion ofand through the puddle clay was the cause of the settlement that led to the

catastrophe. The Home Secretary, Sir George Grey, arranged for Captain R CMoody of the Royal Engineers to inspect the remains of the dam and give expertevidence at the inquest. It is noteworthy that the Bilberry dam had three adversefeatures: a puddle clay filled cut-off trench in which springs were encountered, highlypermeable fill on either side of the puddle clay core, and a culvert through theembankment.

Construction work on the 29-m high Dale Dyke dam started in J anuary 1859. Thedeep puddle clay filled cut-off was completed in 1861 and the embankment wasfinished by April 1863. Impounding commenced in J une 1863 and by 10 March 1864the water level was 0.7 m below the crest of the weir. In the late afternoon of the

following day, a crack was observed along the downstream slope near the crest ofthe dam. At 23:30 a collapse occurred and the dam was breached. In the resultingflood, 244 lives were lost and extensive property damage was caused, includingsome in Sheffield. Robert Rawlinson and Nathaniel Beardmore were appointed bythe Home Secretary, Sir George Grey, to investigate the failure. In their reportRawlinson and Beardmore were critical of both the design and construction of thedam. They believed that failure was most likely to have been caused by leakage froma fractured outlet pipe which passed through the embankment, but the design andconstruction of the embankment itself was also criticised (Rawlinson and Beardmore,1864).

the puddle-wall is much too thin, and the material placed on either

side of it is of too porous a character.No puddle-wall should ever beplaced betwixt masses of porous earth, as puddle, under such conditions,

will crack, and is also liable to be ractured b ressure o water.



19/167

However, the cause or causes of the disaster have continued to be disputed (Binnie,1978). It has, however, been established that the core was susceptible to hydraulicfracture.

Figure 2-3 Postcard photograph of the Dale Dyke dam breach (courtesy of C Hoskins)

As a result of the failures of Bilberry and Dale Dyke and some serious problemsencountered with other embankment dams, important lessons were learned whichled to developments in design and construction during the latter half of the nineteenthcentury. Dale Dyke had a puddle clay filled cut-off trench, permeable fill immediatelyon either side of the puddle clay core, and outlet pipes laid in a trench beneath theembankment. Most attention had always been given to the central vertical core ofpuddle clay which formed the all important watertight element within the embankmentdam. Although the leading engineers of the period realised the need for a substantial

width, in some dams the core was excessively narrow. At Dale Dyke the top width ofthe core was a mere 1.2 m, and, with batters of 1:16, the maximum width at groundlevel was only 4.9 m. Following the failure, the replacement dam constructed nearbyhad a much wider puddle clay core.

According to the earliest concept of puddle clay core dams, the embankmentshoulders merely served to support the core and only needed to be stable andreasonably solid. Captain Moody strongly criticised this aspect of the design. Moodydrew attention to the failure to properly control fill placement and ensure that themore cohesive fill was placed next to the puddle clay core with the more granular fillin the outer slopes. Deficiencies in the supporting fill were also apparent at DaleDyke and it was recognised that it was not prudent to place poorly compacted, highly

permeable fill next to the puddle core. It became recommended practice to place a



20/167

selected fill, which was more cohesive and better compacted than the generalembankment fill, on either side of the puddle clay core.

Captain Moody remarked that a sinkhole in the crest at Bilberry was located abovethe culvert and is no doubt due to the washing away of the bad puddling over andabove the culvert where it passes through the puddle wall below (Moody, 1852).Robert Rawlinson, who had no connection with the design and construction of theBilberry dam or the investigation of the failure, a few years later commented on thedisaster to the effect that The Holmfirth embankment was said to have beenrendered "rotten" by a spring, or springs, in the centre (Rawlinson, 1859). J FBateman, one of the leading dam engineers of the time, claimed that the dam wasbadly constructed on sandstone rock: The water escaped through the fissures of therock, and gradually washed the embankment down in such a way that the top of theembankment was lower than the top of the swallow which was constructed as awaste weir for the purpose of letting the water off (Bateman, 1879). As the dangersof erosion of the puddle clay in the cut-off trench into a fissured rock foundationbecame better appreciated, the superiority of backfilling the cut-off trench with

concrete rather than puddle clay was recognised and there was a general trend frompuddle clay to concrete filled cut-off trenches. Grouting of foundations came into usein the late nineteenth century when Thomas Hawksley applied it to a wing trench atTunstall dam and remedial work at Cowm dam in 1879.

There was a realisation that when puddle clay is in contact with jointed rock,particularly at the bottom of cut-off trenches, water might fracture and erode the clayand escape through joints in the rock. Pentwyn dam (Binnie, 1987a) completed in1863 used a puddle clay cut-off. A fault across the valley and the presence oflimestone resulted in serious leakage from the reservoir, accompanied by settlement.This and other examples of erosion led to the practice of lining the bottom andsometimes the sides of the cut-off trench with brickwork or concrete as was deemed

necessary.

For the second Woodhead dam completed in 1876, Bateman used concrete in thecut-off trench, and this appears to be the first occasion when reliance was placed onconcrete alone. However, use of puddle-filled trenches carried on into the twentiethcentury and incidents of internal erosion continued as at Walshaw Dean Lower andMiddle dams, completed in 1907.

Initially, the practice with puddle clay core dams was to lay the outlet pipe through theembankment and puddle core. In some cases the pipes were surrounded with fillmaterial, but it became more common to surround pipes with concrete. Rawlinsoncame to definite conclusions about the unsuitability of these practices (Rawlinson,

1879).

An engineer, whether designing waterworks or other works, shouldnot put any portion of the material liable to decay out of reach: heshould not bury such material as cast iron under an embankmenthaving a 500-feet base, so that nothing but the destruction of thebank could ever render it accessible for repairs.

The practice of placing outlet pipes, or culverts containing outlet pipes, through theembankment was largely superseded by the more costly but much safer expedient ofdriving a tunnel through the natural ground. When the hazards of a pipe containingwater under reservoir pressure were better understood, the early practice of



21/167

controlling the flow of water by a single valve at the downstream toe was widelychanged and upstream controls became more common.

The problems experienced with puddle clay core embankment dams turned thethoughts of Major Hector Tulloch, who later succeeded Rawlinson as the chiefengineering inspector of the Local Government Board, toward masonry dams and hewrote to Professor Rankine at Glasgow University I consider the fact of the puddlewall in the middle of the dam being virtually all the resistance that the dam can bringto bear against the water, renders all our dams far too weak. In his reply Rankinestressed the importance of the foundation which should be sound rock, ifpracticable, and should a rock foundation be unobtainable, firm impervious earth. Headded that It may be doubted whether any earthen foundation is thoroughly to berelied on where the depth of water exceeds 100 or 120 feet. Rankine also warnedTulloch that It is not advisable to build a masonry dam on an earthen foundation(Tulloch, 1872).

Tullochs misgivings received further confirmation from troubles experienced at

Lower Lliw, a dam designed by Robert Rawlinson, the engineer appointed by thegovernment to investigate the Dale Dyke catastrophe. Construction of the 27-m highdam north of Swansea commenced in 1862 and was completed in 1867. In 1873,water started to flow from downstream drains at a much increased rate and the waterwas turbid. A spring had burst through the puddle clay core. Erosion of the puddleclay led to settlement of the embankment. Remedial work involved an open cutting50 m wide at the top and 15 m wide at the bottom to a depth of 11 m below the top ofthe embankment and a trench nine metres long and six metres wide sunk from thebottom of the cutting to the rock, a total depth of 32 m below the top of theembankment. At a depth of seven metres in the trench, a fissure 0.6 m wide wasfound in the puddle clay filled with the coarse material of the selected fill. The fissureextended down to the face of the rock. A drain was installed to take away the

springwater which acted on the clay at the bottom of the trench and the trench wasbackfilled with puddle clay. In 1883 after two years of service, leakage againincreased. Turbid water came from the drains and settlement occurred at the locationof the remedial works.

Another leading dam engineer, Thomas Hawksley, was called in to advise, but hisanswers to the questions of the dam owner, Swansea Borough Council, were notencouraging. It was not possible to determine the cause of the leak, the materials inthe embankment were fit for purpose and, as to remedial work, he could onlycomment that The method in the former instance was, in my judgement,unexceptionable, and nothing better than a repetition of the same method can nowbe suggested (Binnie, 1981). A disillusioned town council did not opt for a second

attempt at the remedial works and the reservoir was operated from 1883 to 1975 withthe top water level reduced by 5.5 m. It would be nearly another hundred yearsbefore an understanding of hydraulic fracture was gained. The embankment was fullyrebuilt in 1978.

At the inquest following the Dale Dyke disaster, the jury stated that the legislatureought to take such action as would result in government inspection of all works of thischaracter and that such inspections should be frequent and sufficient and regular.However, no legislation was introduced at that time. For the next 60 years there wasno repeat of the major loss of life that had occurred at the two Yorkshire dams. Thecollapse of Cwm Carne in 1875 was the most serious failure (Smith, 1992). This 12-m high embankment settled over many years due to internal erosion and wasovertopped at 17:30 on 14 J uly, with failure at 23:00. The resulting flood caused the



22/167

loss of 12 lives. The failure is reminiscent of Bilberry in that deterioration occurredover a long period with little attempt to improve the obviously defective dam.

The first large masonry dam to be built in Britain was Vyrnwy dam, completed in1891 and designed by Thomas Hawksley and G F Deacon. Following the failure ofHabra dam in French Algeria in 1881, Hawksley became concerned about uplift asan overturning force on the underside of a masonry dam and consequently increasedthe thickness of Vyrnwy. As a further precaution against uplift, a drainage tunnelslightly above the tailwater level was incorporated in the dam, making Vyrnwy notonly the first dam in the world to be designed with uplift acting on its base taken intoconsideration, but also the first to have underdrainage.

2.3 Early twentieth century (1901-1930)

In 1925 two failures caused loss of life: a small dam at Skelmorlie in South WestScotland failed and a disaster at Dolgarrog in North Wales, the latter failure being the

more serious.

Following heavy rainfall, the Skelmorlie lower reservoir failed at 14:00 on 18 April1925. The flood water, which was probably sufficient to overtop the embankment,was augmented by water from a quarry. This quarry had partially filled due to ablocked culvert and suddenly emptied when the blockage cleared. The embankmentbreached over a nine-metre length and the reservoir emptied in 15 minutes. Fivepeople were killed in the village of Skelmorlie. The failure was attributed to a grosslydeficient overflow and inadequate freeboard.

The verdict of the jury at an enquiry held at KilmarnockSherriff Court was: The disaster was caused by absence of

any regular skilled supervision and inspection.

On 2 November 1925, the remote concrete Eigiau dam collapsed, where it was onlyfive metres high, due to a blow-out of the lower part of the dam wall at a point wherethere had been a seepage path for several years. The reservoir water was releasedand surged into the nearly full Coedty reservoir below. The 11-m high Coedty damhad been constructed in 1924 with earthfill shoulders supporting a central concretecore wall. When the dam was overtopped, the material supporting the core wall onthe downstream side was washed away and the core wall collapsed. Sixteen peoplewere killed in Dolgarrog by the resulting flood. Technical evidence at the inquest wasgiven by Ralph Freeman to the effect that the foundation of the dam had not beensufficiently deep. The jury returned a verdict of accidental death caused by the

bursting of the dam under the wall in consequence of the wall lacking a properfoundation. The coroners jury recommended regular government inspection.



23/167

Figure 2-4 The breach in Coedty dam (courtesy of Chris Hoskins)

Only 11 months before the Dolgarrog disaster, Cowlyd dam in the adjacent valleywas nearly breached due to overtopping during a storm on New Years Eve of 1924(Knight, 1975). A V-shaped area of the downstream fill was eroded down tofoundation level, exposing the concrete core-wall. Frenzied backfilling on thefollowing morning saved the dam. Had the central core not been of concrete, it is

possible the dam could have failed leading to certain loss of life. The reservoircapacity was probably more than twice that of Eigiau and Coedty combined.However, floodwaters from Cowlyd would not have gone down the same valley asthose from Coedty and were likely to have missed the main part of Dolgarrog.Subsequently, the spillway crest of Cowlyd was lowered and the wave wall wasraised.

The reservoir failures of 1925 led to the Reservoirs (Safety Provisions) Act 1930.Since this Act was brought into force, and periodic inspection by a qualified engineerbecame mandatory, there have been no dam failures in Britain which have causedloss of life. Although complete failure and breaching of embankment dams has beenrelatively rare in Britain, there have been many serious incidents affecting dams in

service. In some instances these incidents have warranted emergency drawdown ofthe reservoir and costly remedial works or permanent lowering of the top water level.

Whilst after 1900, traditional methods of gravity dam construction continued to beused in which large rocks were laid on mortar or a thin layer of concrete, massconcrete started to replace them in which concrete was placed in thick layers andlarge plums or displacers of rock were embedded. Blackbrook dam, completed in1906, was the first dam constructed using mass concrete with displacers. Theminimal damage during the earthquake incident at this dam illustrates the robustnessof the construction.



24/167

2.4 Mid-twentieth century (1931-1960)

Failures which occur during the construction of embankment dams are generallyassociated with slope instability and are similar, therefore, to the failure of any othertype of earth embankment. The presence and shape of a clay core, especially a wet

puddle clay core, has a considerable influence on the stability of the embankment.The core should consolidate with time and therefore stability should generallyimprove, with the end of construction or the rapid first filling of a reservoir likely to becritical periods for this form of construction. Studies of instability during theconstruction of earth dams have had a major influence on the development of soilmechanics in Great Britain.

During 1937, major slips occurred at three embankment dams under construction.The slip at Abberton took place in J uly and at Hollowell in October. However, the bestknown of the three failures is the instability which occurred at the end of J uly in theearth embankment under construction for the William Girling storage reservoir atChingford in Essex. With eight metres of the planned 10-m height completed, a 90-m

long section of the downstream (outer) slope moved. The embankment had a centralpuddle clay core and was founded directly upon a layer of soft yellow clay. Ageotechnical investigation was carried out by the Building Research Station (Coolingand Golder, 1942). The failure surface passed through the puddle core and thenfollowed a path contained within the layer of soft yellow clay. The undrained shearstrengths of the yellow and puddle clays were measured by laboratory direct sheartests yielding values of only 14 and 10 kPa respectively. A stability analysis wascarried out in terms of total stresses and a factor of safety close to unity wasobtained.

The Chingford reservoir embankment was one of the first to be built in Britain usingwhat at that time would have been described as 'modern earth-moving equipment'.

Thus, the construction rate would have been faster than had previously beencommon practice. It seems likely that the development of high pore water pressuresin the yellow foundation clay due to rapid loading by the embankment was a majorcontributing factor to the failure.

In September 1941 movements were observed in the pitching on the upstream slopeover the central section of Muirhead dam which was under construction in southernScotland. The embankment was 21-m high at this stage and a further five metres offill had still to be placed. The embankment had slopes of one in three, a centralpuddle clay core and shoulders of boulder clay. The Building Research Stationcarried out an extensive investigation of the failure. An initial survey established that

the upstream slope had moved outwards up to 1.2 m and that a berm on thedownstream slope had moved 0.6 m. Movements were horizontal and the toe wallshad not moved. When 0.5 m of fill was added, further horizontal movements of about0.3 m were monitored. It was believed that the embankment had failed through thelower part of the shoulder fill. The strength of this material was found to be veryvariable but the average measured value of undrained shear strength was close to40 kPa which corresponded with limiting equilibrium. The final height of the dam waslimited to 21 m and the upstream slope was stabilised by a substantial berm (Banks,1948).

At the time of the Muirhead failure, a similar embankment was under constructionnearby at Knockendon and the fill had reached about one-fifth of the full height. As a

result of the events at Muirhead, the cross-section of Knockendon was modified byadding a toe weight to the upstream shoulder and by including a zone of stronger



25/167

granular fill in the downstream shoulder. Standpipe piezometers were installed tomonitor construction pore pressures in the fill and check the rate of consolidation.Measured pore pressures were used together with the results of drained shear boxtests to calculate the stability of the embankment (Banks 1952).

The measurement of pore water pressures by the Building Research Station at thesite of the 33-m high Usk dam during the early 1950s had an important influence onthe development of embankment dam design and construction techniques in Britain(Penman, 1978). Usk has a central core of puddle clay and a cut-off trench filled withconcrete. The shoulders of the embankment were of boulder clay. A silt layer wasfound to be present under the downstream shoulder and as a consequence, a sanddrain system was installed. Twin tube hydraulic piezometers were installed in the siltlayer to check on the performance of the drains. Also three piezometer tips wereinstalled at the mid-depth of the first season's fill in the downstream shoulder in J uly1952. The piezometer tips in the silt layer in the foundation measured no significantpore pressures, indicating that the drainage system was effective. However, porepressures in the fill were large. Effective stress stability analyses indicated that the

factor of safety would be unacceptably small if the dam was brought to full height withthe average pore pressure ratio (ru) greater than 0.5. The pore pressure dissipationthat occurred during the winter shutdown period was insufficient to ensure stability.

Advice was sought from Professor Skempton of Imperial College London. Fifteensteel standpipes driven into the fill confirmed the BRS pore pressure measurementsand it was decided to place horizontal drainage layers within the embankmentshoulders: it is believed that this was the first use of drainage blankets of this type inan earth dam to control construction pore pressure (Sheppard and Little, 1955). Theuse of instrumentation and associated construction techniques such as thosedescribed above have considerably reduced the risk of embankment failure duringconstruction.

Slope instability is by no means confined to the construction period. A majordownstream slip was discovered on 18 December 1951 at Harlow Hill, anembankment dam forming an open service reservoir which had been built atHarrogate in 1868. The slip occurred following an extremely wet autumn. A verticalmovement of 0.3 m had occurred on the slip plane adjacent to the puddle clay core,with 0.23 m uplift at the toe against the concrete retaining wall. Movement wascontinuing at 0.01 m per hour. There was a clear danger of a catastrophic dambreach and emergency actions were taken: the reservoir was lowered as fast aspossible; sandbags were placed on the toe of the slip to improve stability; andtarpaulins were placed to prevent further ingress of rainfall into the embankment.Movement monitoring commenced and the police were alerted to be ready to

evacuate the downstream population. The 1:1.9 slopes of the embankment were toosteep for the clayey embankment fill and slope stability must always have beenmarginal. The embankment was therefore a disaster waiting to be triggered by somephenomenon such as unusually heavy rainfall.

Some important lessons were drawn from the incident (Davies, 1953); the mostimportant was that all embankment dams of clay constructed before the advent ofsoil mechanics should be regarded as suspect. Furthermore, the normal visualinspection of a dam, unsupported by any real knowledge of the properties ofmaterials of construction, is insufficient to determine the stability of the structure. Soilstrength tests are essential to determine actual stability, but they must be sufficientlynumerous to provide a proper statistical average, and must be taken from locationson potential slip planes. However, soil strength information is currently only availablefor a minority of Britains embankment dams.



26/167

Despite the clear warnings given at the time, the lessons of Harlow Hill were notgenerally learned. However, corroboration of the likelihood of some old embankmentdams having only marginal slope stability came when a comprehensive programmeof investigations and remedial works was undertaken in the 1970s in NorthernIreland. Most of the seventy or so reservoirs in the province were impounded byembankment dams and at nine of these, it was necessary to install filters withoverlying rockfill stabilising berms (Cooper 1987).

Slope stability failures were by no means the only type of incident to afflictembankment dams in this period. One particular type of problem was associated withthe precautionary partial drawdown of reservoirs during the Second World War. The7.6-m high King George V reservoir near Chingford consists of clay fill with a puddleclay core. The embankment was built on unstripped grass and topsoil without anyspecial provision for underdrainage of the downstream (outer) slope. In September1939 is was decided to reduce the top water level by 1.5 m and this restriction wasmaintained until February 1945 when raising the water to its previous top level

began. Water leakage appeared at the toe of the embankment as the original topwater level was approached. The possibility of the leakage indicating an incipientmajor failure was recognised and the water level was lowered. An investigationinitiated in association with the Building Research Station revealed the presence ofroots in the puddle clay down to the previous temporary top water level 3.1 m belowthe crest of the bank. This evidence together with extensive field observations andsoil testing indicated that the passage of water was through the upper part of thepuddle clay core which had been subject to drying, shrinkage and cracking during theyears 1939-1945 (Bishop, 1946).

Refilling of a long-empty reservoir impounded by an embankment dam having apuddle clay core should always be undertaken with the utmost caution due to

possibility of desiccation and cracking of the upper part of the core. Additionalsettlement resulting from the major drawdown could lead to water flow over the top ofthe core. The possibility of hydraulic fracture during refilling should also beconsidered. Caution is necessary in the case of a dam having any or all of thefollowing features: (a) a puddle filled cut-off trench, (b) permeable fill on either side ofthe puddle clay core, and (c) outlet works passing through the body of theembankment.

Earth dams may fail due to inadequate spillways when an exceptionally large floodoccurs, however the damage caused by a severe rainfall event may be considerableeven without a dam failure. On 15/16 August 1952, 230 mm of rain fell in 24 hours onthe upper valleys of the East and West Lyn rivers in Devon. At 20:30 on 15 August

1952 the services of the fire brigade were requested above Barbrook on the WestLyn where a dam had burst and flooded Radsbury Farm (Delderfield, 1981). Withsuch immense rainfall the failure of a small dam was largely irrelevant in thesubsequent catastrophe in which the West Lyn river burst its banks and a torrentswept through the town of Lynmouth, resulting in 34 deaths and making a thousandpeople homeless.

Earthquake damage to British dams is comparatively rare, but on 11 February 1957Blackbrook concrete and masonry gravity dam was affected by an earthquake withlocal magnitude 5.3. The dam suffered fairly superficial damage includingdisplacement of 0.75 tonne copings and manhole covers which sheared and were

displaced up to 20 mm. However, it has been asserted that Had the line of the



27/167

tremor been 90 degrees displaced the result would have been catastrophic as thedam is located only four miles north of the epicentre (Kennard and Mackey, 1984).

2.5 Late twentieth century (1961-2000)Some of the most troubling incidents in this period have involved internal erosion andsuch problems were not confined to old puddle clay core dams. On first impoundingin 1966, just before the reservoir was full, the main underdrain flow at the newly built48-m high Balderhead dam increased. The dam has a rolled boulder clay core,relatively stiff shale fill shoulders and a concrete cut-off. The top 10.8 m of the claycore has vertical sides. Immediately downstream of the core is a crushed limestonefilter which connects with the ground drainage blanket. The filter and the drainageblanket were designed according to standard filter rules. Subsequently, localisedsettlements occurred along the crest and in 1967 two sinkholes formed in the crest.The reservoir was immediately drawn down by 9.2 m and the underdrain flow

returned to its previous level. It was established that the main underdrain flow hadturned cloudy about a month before the first sinkhole appeared, but after drawdownthe water became clear.

Exploratory boreholes revealed erosion within the core at several locations; theboulder clay material had become segregated and the finer particles lost by watererosion. The damage was associated with cracking which had been initiated byhydraulic fracture of the core under almost full reservoir pressure. Low stresses in thecore were caused by arching between the clay core and the shoulders and possiblyby longitudinal strain due to differential settlement across foundation discontinuities.It was also postulated that once the cracks had formed they were kept open by thewater pressure and under the low flow conditions the coarser eroded material hadsegregated in the cracks. On drawdown the seepage paths closed up due to thedecrease in water pressure. Over the central 200 m of the dam, covering the zonesof worst damage, the core was repaired by constructing a 0.6 m wide diaphragm walldown to the concrete cut-off. As well as serving as an additional water barrier, thediaphragm wall was intended to prevent migration of eroded material through thecore (Vaughan et al., 1970).

The internal erosion problems at Balderhead led to major investigations and researchby Professor Peter Vaughan at Imperial College London, which has not only built amuch better understanding of the mechanisms involved in the internal erosionprocess, and particularly the role of hydraulic fracture, but has also led to important

developments in filter design (Vaughan and Soares, 1982).

On 23 December 1969 a horse fell into a two-metre deep hole in the crest of the 24-m high Lluest Wen dam in South Wales, which had been built in 1892. Subsidencehad occurred previously in 1912 and 50 tonnes of cement grout had been injected inthe area of the valve shaft in 1915-16.It was feared that the dam would collapse andan emergency was declared by George Thomas, the Secretary of State for WelshAffairs. The infirm and elderly were evacuated from their homes on the night of 12/13J anuary 1970 (Gamblin and Little, 1970). The 0.38-m diameter draw-off pipe wasinadequate for rapidly lowering the reservoir water level and a large number ofpumps, some positioned by helicopter, were brought in to lower the water level. Also,an emergency cut was made through the spillway lowering the overflow level of the

reservoir. The reservoir level was lowered by 9.1 m in twenty days (Twort, 1977).



28/167

Meanwhile, over 18 tonnes of clay/cement grout were injected into a single hole inthe neighbourhood of the valve shaft where subsidence had occurred.

With the emergency over, full grouting of the puddle clay core was done and fiftytonnes of clay/cement grout were injected. After completion of the grouting works,further investigation was carried out. The puddle clay core was found to consist ofsandy silty clay with pockets of silt and sand. The core had a series of cracks, manyof them open and iron-stained by seeping water. About 75 per cent of the crackswere within five degrees of the horizontal. Open water-worn cavities were found intwo drill-holes. The water content and undrained shear strength of the puddle clayfluctuated widely and erratically. The core was very soft in the vicinity of the valveshaft. In view of these findings, it was decided that grouting alone could not provide asatisfactory solution and a new plastic concrete core was installed using the slurrytrench method.

During the excavation of a six-metre diameter shaft at the valve tower, it wasdiscovered that the brickwork of the draw-off tunnel had not been bonded into the

masonry at the back of the valve shaft. Puddle clay had eroded through a 50-mm gapand then through a crack in the 0.15-m diameter pipe. At the time of the emergency,there was a 0.06 cubic metre pile of puddle clay at the downstream end of the 0.15-mdiameter pipe. It was of concern that so much hinged on a tiny detail which mightnever have been detected until perhaps too late but for the requirement for majorremedial works. The extreme seriousness with which the incident was viewed, andthe emergency measures put in place by the Welsh authorities, were undoubtedlyinfluenced by the Aberfan disaster three years earlier, when on 21 October 1966,following heavy rainfall, a colliery spoil tip collapsed and 150,000 m3 of spoil floweddownhill into the mining village, killing 144 people, 116 of them children (Bishop etal., 1969).

Warmwithens dam was a 10-m high clay fill embankment built more than 100 yearsago near Oswaldtwistle in Lancashire. The reservoir it impounded lay in series abovetwo other small reservoirs: Cocker Cobbs and J ackhouse. During the period 1965 to1966, the dam was raised to provide adequate freeboard and the old cast iron draw-off pipe was replaced by a reinforced concrete segmental tunnel driven through theembankment. The tunnel contained a steel pipe for the water outlet. At 7:30 on 24November 1970 an escape of water was detected and by 13:30 the dam wascompletely breached to foundation level (Wickham, 1992). The water impounded bythe dam was discharged into the two lower reservoirs. The embankment dam ofCocker Cobbs was overtopped, but it did not fail, and the water passed the spillwaysof the lowest reservoir, J ackhouse, without causing serious damage. Had a cascade

failure of the two lower dams taken place, the resulting flood could have causedserious damage in Oswaldtwistle. The breach occurred along the line of the outlettunnel. It therefore seems possible that seepage through or along the perimeter ofthe abandoned cast draw-off pipe, or along the perimeter of the new tunnel, couldhave played a part in causing the failure. This incident showed how rapidly aninternal erosion incident can develop and confirmed the hazard where a structurepasses through the clay core of an embankment dam.

On the afternoon of 7 March 1983, a member of the public taking two dogs for a walknoticed a depression in the asphalt of the crest roadway of the 35-m highGreenbooth dam about 20 m from the west abutment. The dam, built near Rochdale,was completed in 1962 and was one of the last dams to have a puddle clay core. The

depression deepened quickly over a few days. By mid-morning the next day itmeasured three metres by one metre in plan and had subsided by 0.16 m. The



29/167

depression was directly above the toe of a concrete wing wall where there was asharp change in direction of the concrete/puddle clay core interface. A panelengineer was appointed to supervise the investigation and the reservoir level, whichwas 1.65 m below top water level, was reduced by 9.3 m over an eight-day period.An investigation identified voids in the puddle clay. These were grouted by tube--manchette used a specially designed bentonite, cement, fly ash, clay grout(Flemming and Rossington, 1985).

Serious incidents have not been confined to reservoirs that came within the ambit ofthe Reservoirs Act 1975. Since the provisions of the Act only applied to reservoirsimpounding in excess of 25,000 m3 above natural ground level, the Act did not applyto a planning application submitted for a 14,000 m3 pond at Horndoyne farm nearAberdeen in March 1989. Impounding took place in late autumn 1990 and theembankment dam breached during the night of 17-18 November 1990. Water hadbeen seen to trickle along the side of the outlet pipe and this developed into a streamtaking earth with it. Eventually a breach was formed and a wall of water, a metre ormore deep, swept down the small valley. Four houses were flooded causing

considerable damage to the buildings and their contents. A large residential caravanwas swept over 100 m from its site, but there were no injuries to people. The failureillustrated the dangers posed by small reservoirs outside current reservoir safetylegislation. Measurements made subsequent to the failure suggested a likely storagecapacity of 23,000 m3, half as much again as the approved scheme and close to the25,000 m3 threshold for the provisions of the Reservoirs Act to then apply.

Advances in soil mechanics through the 1960s should have greatly reduced thepossibility of instability during embankment construction, but this type of failure wasnot eliminated as the major upstream slip at Carsington in 1984 demonstrated. At thebeginning of J une 1984, a 400-m length of the upstream shoulder of theembankment dam slipped some 11 m. At the time of the failure, embankment

construction was virtually complete with the dam approaching its maximum height of35 m. Horizontal drainage blankets were incorporated in both the upstream and thedownstream shale fill shoulders. Piezometers had been installed and pore pressureswere being monitored in the foundation, in the clay core, and in the shoulder fill.Effective stress stability analyses had been carried out. The failure surface passedthrough the boot shaped rolled clay core and a relatively thin layer of surface clay inthe foundation of the dam. Investigation of the events at Carsington has madeimportant contributions to the fundamental understanding of the behaviour of largeearthworks of this type (Vaughan et al., 1989; Dounias et al., 1996).

In his report to the Secretary of State for the Environment on the Carsington failure,Roy Coxon made a number of recommendations including the use of review or

advisory panels for major dam construction projects (Coxon, 1986).

There is merit in involvement of a Board or Panel of Specialists in projectsof this kind to review key elements relevant to design and construction. Sucha Board can in no way relieve other parties of their normal responsibilities.

This sensible recommendation, which follows international good practice, has beenfollowed on a number of new dams and major works, including the reconstruction ofCarsington, the construction of Queen's Valley dam, J ersey (1986-1993),modification of Woodhead dam (1988-1991), rebuilding of Audenshaw No 3 reservoir(1988-2002) and the raising of Abberton dam (2008-present).



30/167

A wet spot on the downstream slope of Lambieletham dam, near St Andrews, wasobserved during routine surveillance in November 1984. Within 24 hours seepagehad increased and muddy water was observed at the base of an eight-metre longcrack, which had a maximum width of 0.4 m. There was evidence of uplift six metresdownstream of the crack. Following inspection by a panel engineer, it was decidedthat the reservoir should be emptied as quickly as possible. On the night of 20November, engineers and the police assessed the likely consequences of failure andhouseholders in the area were alerted to the situation. Pumps were brought onto thesite by helicopter and the reservoir level was lowered by five metres in three days.The dam was demolished in October 1985 and BRE carried out extensiveinvestigations as fill was excavated. It was concluded that downstream slopeinstability was triggered by high pore water pressures associated with large volumesof water from the north-west valley side flowing into and saturating the lower half ofthe downstream shoulder fill.

Other types of serious incidents at embankment dams have involved damage to theupstream slope protection due to wave action. As a result of these incidents, work on

wave prediction has been carried out by Hydraulics Research and cases of failurehave been evaluated at dams with three types of upstream protection: pitching,concrete blockwork and concrete slabbing. On the basis of this research, guidancehas been produced on best practice in the design of upstream slope protection(Herbertet al., 1995).

In some cases, wave action under storm conditions is so great that downstreamslope stability is affected. In February 1962, a major storm at Blithfield reservoircreated severe wave action that overtopped the dam, saturated the downstream filland caused a slip in the downstream shoulder (Leach, 1975). The downstream slip atCombs in J anuary 1976 also occurred during a storm and was probably triggered bywave action saturating the downstream fill through cavities in the wave wall

(Ferguson et al., 1979).

Two serious incidents at service reservoirs are worthy of note, although neitherincident led to a catastrophic release of the reservoir water. The roof of Sheephousesreservoir, near Bacup, consisting of pre-stressed concrete beams collapsed in 1962.The failure of the beams has been attributed to a reduction in the strength of thehigh-alumina cement concrete (Neville, 2009). In October 1979, a suddensubsidence occurred in the south-west corner of the No 1 Mill Hill service reservoir,which is built on a limestone foundation. Part of the structure collapsed and thedivision wall was also affected. Water stored in the damaged compartments drainedinto the subsided area and then into underlying strata (Millmore and Heslop, 1988).

Although incidents at concrete dams have been relatively rare, major investigationsand remedial works have been required at a number of such dams, often associatedwith uplift pressures not being allowed for in the original design of older dams or withconcrete deterioration. A statutory inspection of the Carron dam confirmed doubtsabout the stability of the concrete gravity section and the structural inadequacy wasremedied by installing pre-stressed rock anchors to increase the factor of safetyagainst overturning (McKenna, 1996; Kennard et al., 1996). At Upper Glendevon thedam was strengthened by the addition of downstream rockfill (J ohnston, 1995). AtArgal dam, there was concern about the condition and performance of post-tensioncables which had been installed during raising works. This led to an investigationwhich included deformation monitoring using electro-levels (Tedd et al., 1995). Aconcrete buttress was subsequently placed on the downstream side.



31/167

Stability checks on gravity dams in the north of Scotland identified three masonry andconcrete gravity dams where the stability under the design flood was not satisfactory.A fourth dam, Mullardoch, a 48-m high mass concrete gravity structure, requiredmeasures to be taken to improve stability due to a developing situation (Peacock andSandilands, 1993). It was reported that on 4 J uly 1986 leakage at Mullardoch hadincreased to 5.2 litres per second from 0.16 litres/second on 25 J une. Existing crackshad opened up and there was some evidence that uplift pressures had increased(J ohnson, 1986). Rock anchors were installed to overcome concerns about crackingand leakage (Gosschalk et al., 1991).

It

charles a-tedd p-warren a-lessons historical dam incidents-2011

Documents